Re: compiling glibc-2.1.3

["Christopher C. Chimelis" <chris debian org>]

On Mon, 13 Mar 2000 cpg@aladdin.de wrote:

> I've tried to compile glibc-2.1.3 on my AlphaStation 200 (with gcc-2.95.2)
> and I get the
> following error when it tries to compile  math/conj.c:
> 
> ----------------
> In file included from ../include/bits/cmathcalls.h:1,
>                  from ../math/complex.h:88,
>                  from ../include/complex.h:1,
>                  from conj.c:21:
> ../math/bits/cmathcalls.h: In function `cimagf':
> ../math/bits/cmathcalls.h:139: Unable to access imaginary part of complex
> value in a hard register on this target
> make[2]: *** [/usr/oth/tmp/glibc-2.1.3-build/math/conj.o] Error 1
> ----------------
> 
> This is a 21064 machine, maybe that's the reason?
> Anybody knows a solution?

This happens because of the lack of complex support in gcc 2.95.x.  The
older egcs 1.x.x versions work, but any complex operations will return
bogus results, so the gcc group figured that, until they implemented
proper complex support on Alpha for gcc, they would just not have it do
anything at all (which I agree with).  Since then, a proper patch has been
applied to the snapshot tree supporting complex types more fully.  In
short, you'll have to use a relatively new gcc snapshot (past few months
should do) if you want to compile glibc 2.1.3.  Either that or revert to
an old egcs (if you don't plan on needing complex support to actually
work).

C